One of the organizational benefits is to reduce need for network servers within an organization, thereby causing less IT headache. Users don't really know where their files are anyway, so switching to the cloud from a network share shouldn't be a problem. In the future, storage space, bandwidth, and application usage may be measured and billed like electricity or phones are now.
The federal government's CIO, Vivek Kundra, is looking to the cloud as a solution to the challenge of spiraling IT infrastructure costs. While acknowledging a transition period, he seems to believe that the cloud will result in greater efficiencies and huge cost savings. From what I've read, he bases his argument on his work with the Washington, DC government, where he introduced Google applications to over 30,000 government workers. Reportedly, over 4,000 (about 13%) are using the cloud on a regular basis. The federal plan is to roll out similar cloud services to 300 million worldwide government workers. With similar usage rates, there should be about 39 million regular federal cloud users within a year of the official rollout.
What I wonder is, can this cloud experience be replicated on such a massive scale? Whenever I am presented with numbers, I like to do some math. 300 million is 10,000 times 30,000. That's four orders of magnitude from DC government to the federal level. I don't think it's reasonable to expect the tech czar to recreate his success 10,000 times over, but let's look a little further before passing judgement.
To get another perspective, I googled "30,000." Here are some results from the first ten (out of 99,500,000):
- a Washington Post article mentions that 30,000 jobs were created or saved due to the federal economic stimulus package
- the New York Daily News reports that 30,000 poll workers may not get paid for election work in the NYC area this year.
- 30,000 people a year wake up during surgery, according to msnbc.com
- a datacenterknowledge.com interview reveals that Facebook has 30,000 servers
The last two items seem particularly interesting. 30,000 patients seems like a ghastly amount of people waking during surgery, but out of 20 million surgeries in the US each year, maybe it's not so many. After all, 99.85% of patients don't wake up during surgery. Then again, if it were the fabled "5 nines" (99.999%, or down only about five minutes every year) of the IT server world, only 200 patients would wake up. If server uptime was 99.85%, it would be down for an entire day every week. OK, maybe that's an unfair comparison. Anesthesia has only been in regular medical use for about a hundred years; servers have been around for... wait a minute.
That brings us to the item about Facebook. Notice how the same numbers pop up: 30,000 servers, 300 million users, so that's 10,000 users per server. So what? Well, Facebook was down this week for hours, at least for some people. Will the government launch its cloud initiative with 30,000 servers? If only 13% of government workers use the cloud, we still need 3,900 servers to accommodate them, and for more activity than Mob Wars and status updates. Earlier this month, T-Mobile and Microsoft destroyed data for all of its Sidekick phone users, who lost all of their contacts and other personal information. What assurance will the government have that the cloud will be properly maintained and backed up?
The federal government has not had the best track record at massive IT upgrades. Search online for upgrade problems with NMCI or FBI and you'll find a litany of lost money and productivity. The cloud transition will require many of the improvements in government infrastructure that would be necessary without the cloud. Firewalls will need to be modified, browsers upgraded, bandwidth widened, etc. etc. All this has to be done by a group of internal IT folks who are often contractors with a vested interest in the status quo. Even the government employees in IT face reduction in force or at least a transfer.
Maybe a good first test is to push SharePoint into the cloud. Agencies have only been using it for a few years, so it shouldn't have such a backlog of files and legacy applications like so many other servers do. Users don't know where the SharePoint server is, and it doesn't matter anyway, since it is designed for external as well as internal access. Development of most SharePoint applications takes place through the browser or the free SharePoint Designer; there's very little need to login to the server directly. Since many government workers use some form of Microsoft Office in their daily work already, workflows can stay the same, and agencies can avoid supporting a melange of office productivity tools.
Though cloud computing is indeed what the future holds, it is just getting started in the corporate world. It's a good idea to start with technology that is well-suited to the cloud, but familiar to users at the same time. Pushing SharePoint to the cloud first will help ensure a smoother transition.
No comments:
Post a Comment